Live AI Competition Board
AI Race 2026
Frontier model competition board: benchmark score, pricing pressure, and deployment profile in one quick read.
Tokens served today
4.82B
+7.4%
Active users
12.5M
+2.3%
Models deployed
847
+0.6%
Labs tracked
7
+0.0%
Top quality
96.7 GPT-5.5
OpenAI
Top HumanEval
99.5 GPT-5.5
OpenAI
Top MATH
97.9 GPT-5.5
OpenAI
Snapshot
Top now by MMLU + telemetry
OpenAI
96.7 · GPT-5.5
tags Reasoning · Code
Lead span 16.7 · rank window
Google DeepMind
94.1 · Gemini 3.1 Pro
tags 1M Context · Multimodal
Lead span 14.1 · rank window
xAI / Grok
93.5 · Grok 4.20
Lead span 13.5 · rank window
DeepSeek
90.8 · DeepSeek R1
tags Open Source · Math
Lead span 10.8 · rank window
Cheapest in
Price pressure lane
Google DeepMind
$0.00 / $12.00
IO ratio 1200.0x · best in-score lane
Cheapest model spread $12.00
Meta
$0.10 / $0.49
IO ratio 4.9x · best in-score lane
Cheapest model spread $0.39
OpenAI
$0.15 / $15.00
IO ratio 100.0x · best in-score lane
Cheapest model spread $14.85
Mistral AI
$0.25 / $9.00
IO ratio 36.0x · best in-score lane
Cheapest model spread $8.75
Live health
Runtime pulse
286ms · 99.78% uptime · 6809rps
270ms · 99.85% uptime · 2008rps
129ms · 99.67% uptime · 7275rps
Quality leaders
AI quality index top
#1 GPT-5.5
97.2
OpenAI · 1Mctx · launched Mar 2026
Speed 95 · Context 1M · IO 6.00x
Quality 97.2 · Value 98.7 · IO 6.00x
Rank 1 · open no
Vision · Tools · Streaming
#2 Llama 4 Scout
89.8
Meta · 1Mctx · launched Apr 2026
Speed 125 · Context 1M · IO 3.00x
Quality 89.8 · Value 89.2 · IO 3.00x
Rank 2 · open yes
Vision · Tools · Fine-tune · Streaming
#3 Gemini 3.1 Pro
89.7
Google DeepMind · 1Mctx · launched Feb 2026
Speed 85 · Context 1M · IO 6.00x
Quality 89.7 · Value 91.2 · IO 6.00x
Rank 3 · open yes
Vision · Tools · Streaming
#4 Llama 4 Maverick
89.1
Meta · 1Mctx · launched Mar 2026
Speed 120 · Context 1M · IO 2.58x
Quality 89.1 · Value 88.2 · IO 2.58x
Rank 4 · open yes
Vision · Tools · Fine-tune · Streaming
#5 Claude Haiku 4.5
89.1
Anthropic · 200Kctx · launched Jan 2026
Speed 98 · Context 200K · IO 5.00x
Quality 89.1 · Value 88.4 · IO 5.00x
Rank 5 · open no
Vision · Tools · Streaming
Open-first lane
Open models by MMLU
Gemini 3.1 Pro
94.1
ctx 1M · lat 145ms
IO 6.00x · 1M Context · Multimodal
DeepSeek R1
90.8
ctx 128K · lat 95ms
IO 3.98x · Open Source · Math
Llama 4 Maverick
89.8
ctx 1M · lat 90ms
IO 2.58x · Open Source · MoE
DeepSeek V4
89.0
ctx 128K · lat 115ms
IO 1.50x · Open Source · MoE
Value #1
GPT-5.5
OpenAI · 1M ctx
2.5 in / 15 out · 1,000,000 ctx
API model · streaming yes · speed 95
97.2
96.7
98.7
Speed 95 Ctx 1M Value/Eff 39.5
IO 6.00x · Value 98.7
Vision · Tools · Streaming
Latency band 19x
launched Mar 2026
Value #2
DeepSeek R1
DeepSeek · 128K ctx
0.55 in / 2.19 out · 128,000 ctx
Open model · streaming yes · speed 35
84.1
90.8
93.2
Speed 35 Ctx 128K Value/Eff 169.4
IO 3.98x · Value 93.2
Tools · Fine-tune · Streaming
Latency band 7x
launched Jan 2025
Value #3
Gemini 3.1 Pro
Google DeepMind · 1M ctx
2 in / 12 out · 1,000,000 ctx
Open model · streaming yes · speed 85
89.7
94.1
91.2
Speed 85 Ctx 1M Value/Eff 45.6
IO 6.00x · Value 91.2
Vision · Tools · Streaming
Latency band 17x
launched Feb 2026
Value #4
Kimi K1.5
Moonshot AI · 256K ctx
0.5 in / 2 out · 256,000 ctx
Open model · streaming yes · speed 65
86.0
87.3
90.5
Speed 65 Ctx 256K Value/Eff 181.0
IO 4.00x · Value 90.5
Vision · Tools · Streaming
Latency band 13x
launched Jan 2025
Value #5
Claude Sonnet 4.6
Anthropic · 1M ctx
3 in / 15 out · 1,000,000 ctx
API model · streaming yes · speed 75
87.4
90.2
90.3
Speed 75 Ctx 1M Value/Eff 30.1
IO 5.00x · Value 90.3
Vision · Tools · Streaming
Latency band 15x
launched Feb 2026
Value #6
Claude Opus 4.6
Anthropic · 1M ctx
5 in / 25 out · 1,000,000 ctx
API model · streaming yes · speed 50
83.4
90.5
89.7
Speed 50 Ctx 1M Value/Eff 17.9
IO 5.00x · Value 89.7
Vision · Tools · Streaming
Latency band 10x
launched Feb 2026
Speed lane
Llama 4 Scout
125
Llama 4 Maverick
120
Claude Haiku 4.5
98
Context lane
GPT-5.5
1M
Gemini 3.1 Pro
1M
GPT-5.4
1M
Cheapest input
Llama 4 Scout
$0.1
Qwen 3.5 9B
$0.1
GLM-5
$0.15
HumanEval lane
GPT-5.5
99.5
DeepSeek R1
92.5
Mistral Large 2
92.0
Kimi K1.5
89.2
Math lane
GPT-5.5
97.9
DeepSeek R1
97.3
Kimi K1.5
95.8
Claude Opus 4.6
95.0